The pros and cons of labelling animal behaviour.

Many animal behaviour consultants abhor labels.

They consider them not just pointless, but disastrous, and many of them wouldn’t be caught dead using labels.  

You might think I’m exaggerating for dramatic effect, and yes, I do have a penchant for hyperbole so it is entirely possible… but sometimes I do wonder. 

As an ethologist, I had merrily been using labels for decades without even realizing that they could be problematic. It was not until I started hanging out with behaviour analysts that it was pointed out to me.

I had two main reactions to that insight:

• Wow, it’s really useful to realize that labels can be very detrimental!
• Wow, some people really don’t seem to realize how useful labels can be!

When are labels useful? Well, as is the case with literally everything related to animals and their behaviour, it depends on the context.

Labels are sometimes useful, sometimes irrelevant, and sometimes harmful.

I can think of three useful types of label, and one label type that is harmful. And yet, the harmful labels are getting all the attention!

Hence, this blog post.

And I realize that I might have lost some readers already, wondering what “using labels” is all about and why it could potentially be problematic to use them. Or not.

So let’s start with some useful labels.

How do ethologists use labels?

As ethologists, we observe animals to figure out what they’re doing. We might be trying to get a handle on the activity budget of capuchin monkeys, for instance – what a typical day looks like, behaviour-wise.

So, say you’re the one looking at the monkeys, and you need to somehow take notes on what they’re doing, perhaps by whispering a continuous narrative onto a recorder, or perhaps by jotting down what they happen to be doing exactly every five minutes, or whether a specific behaviour was seen within the last 10-minute interval. There are different ways of collecting behavioural data – but we typically end up using some type of shorthand to describe a behaviour, or a cluster of behaviours. All the while trying to ignore the neck pain and the buzzing of the mosquitoes, distracted by thoughts on whether that malaria prophylaxis regime is still up to date…

In ethology, we call these clusters of behaviours behavioural systems.

And essentially, behavioural systems are labels.

Note that these clusters of behaviours, these behavioural systems (labels!) won’t be haphazard and made up in the moment, but rather chosen from some sort of ethogram, from more or less extensive lists of well-defined and species-specific responses.

Affiliative (pro-social) behaviour, for example, might involve the sub-categories “play”, “mutual grooming”, “parenting behaviour” and “sexual behaviour”.

And within each of these sub-categories we might find a number of specific responses. “Play” for instance, might include a number of play vocalizations, meta signals (species-specific play invitations), chasing, changing directions, frolicking, lunging, pouncing, wrestling and biting different body parts. Also, we might note who is doing the chasing, pouncing, biting, and whether biting occurred to a neck or a limb – and who is getting chased, pounced, bitten, as well as the location of the bite – and how and when these roles are reversed.

But when we’re standing there in an awkward position, staring at the monkeys and itching our fresh mosquito bites, we might not care whether the animal was lunging or pouncing or biting the left arm, and instead we’ll simply note the sub-category, or even the behavioural system that we’re interested in. We might note that they’re playing, or affiliating, for instance.

Once we summarize our findings, we might be able to say things like: among tufted capuchin monkeys, juvenile males spend 14,6% of their time playing, whereas juvenile females only play 7,1% of the time (and in case you’re wondering, yes – these are real data).

In other words, in ethology it is part of the job description to use labels – but these labels will typically be a shorthand description of a cluster of well-defined responses shown in certain contexts and with similar function.

People on the Behaviour Analytical side sometimes disparagingly refer to the ethological perspective as etho-babble, suggesting that the labels ethologists use are unfalsifiable, and therefore of no value.

Not only do I disagree with that conclusion, I would go as far as to say that if we were to discuss activity budgets without using labels, we would get a list of thousands of well-defined responses of varying frequency, and seeing any patterns among those behaviours might be really difficult, except perhaps that we might note that chasing is often, but not always, followed by pouncing. In other words, we would likely notice that behaviours occur in clusters, but since we’re not labelling those clusters, I would argue that it would be difficult to make sense of the data.

It would be hard to see the forest for all the trees, as it were.

So, in the field of ethology, we use labels to make sense of behaviour. We use labels to simplify, find important patterns and summarize. A few examples:

• Abnormal behaviour is typically diminished when we scatterfeed food to captive monkeys rather than offering food in bowls.
• Kittens who receive positive human interaction more than 60 minutes per day between the ages of 2-8 weeks become more affiliative as adults than if they receive less than 15 minutes of human interaction.

In these examples, “abnormal behaviour”, “positive human interaction” and “affiliative” are labels – and those labels are useful. We can use the labels to quickly identify areas within animals’ behavioural profiles where we can do interventions that will improve their quality of life.

Without those labels, we would have a haystack of behavioural observations but no obvious way of making sense of them.

So, if labels can be this useful, where does the stark aversion amongst behaviour consultants to use them come from?

Seems to me that the problem is multi-faceted, and there are at least four potential pitfalls when labelling behaviour.

For starters, the labels that non-ethologists use (and the ones that I think behaviour analysts primarily are exposed to) are typically not based on a previously agreed-upon ethogram made up by verbs, but rather, they’re often adjectives with negative connotations, like for instance “stubborn”, “lazy”, or “stupid”.

Problem one: Lack of Clear Definitions

One of the most obvious problems with labels is that different people interpret them differently.

When asked to describe what types of behaviours “stubborn”, “lazy” or “stupid” individuals would show, we might get as many answers as the number of people we asked.

In other words, people don’t agree on what those labels mean.

The connection has been lost between the label and the actual observable behaviours – not to mention the function and context of those behaviours.

If someone says ”he’s dominant” – for them that might mean that their dog walks through the door first.

And here I’ll just have to make a short digression, because this has been bothering me for some time.

Digression on the topic of dominance

In ethology, the label “dominant” refers to an individual who has priority of access to resources relative to another, well-known individual. This is an observable and measurable phenomenon, it’s not a personality trait and has absolutely nothing to do with whoever walks through the door first.

However, many lay people don’t use this ethological definition when it comes to animals, but rather the sociological definition of the label “dominant” – which refers to an individual (human) controlling the behaviour of another individual (also human). And they use this as an excuse to control the behaviour of their animals, often using force.

Let me spell it out: dominance in a group of animals is primarily about priority of access to resources, not controlling the behaviours of others. In fact, it’s often the subordinate individuals in a group that maintains the hierarchy by unobtrusively moving aside when the dominant individual approaches, in order to avoid escalation or conflict. In ethology, we call this unprovoked submission or displacement behaviour (since it’s about being displaced from one location to another – and yes, it’s a confusing label because displacement behaviour can also refer to certain stress-related behaviour such as self-preening).

In general, the presence of an established dominance hierarchy in a stable group means that aggressive behaviour is minimized: most often there’s limited need for it since the subordinate animals tend to yield resources to the dominant – often without them even showing low-level aggressive behaviour.

That’s not to say that rank-related aggressive behaviour is never shown within existing groups of animals; it is. During group composition changes or rank challenges we’ll often see contact aggression resulting in injury. Once rank is established, ritualized threat displays or non-contact aggression is typically more common. However, those aggressive behaviours typically constitute just a fraction of all dominance interactions – but unless you know what to look for, you’ll miss the subtle avoidance behaviours shown by the subordinates.

In short, the existence of dominance hierarchies amongst animals living in stable social groups reduces the costs that would be involved if the animals had to fight over every single resource. Those costs incluce the risk of injury, reduced vigilance and therefore increased risk of predation, as well as energy- and time investments in lengthy and repeated aggressive bouts that could rather be spent on the other activities such as foraging, playing, grooming and mating.  

Again, dominance in ethology – among animals – is not primarily about controlling the behaviours of others.

So, from my perspective, trying to control the behaviour of a dog or a horse through force, and claiming that one is simply “mimicking natural animal dominance behaviour”, is faulty: such forceful approaches would be using the sociological rather than the ethological definition of the term, and a more correct label of that type of behaviour when directed towards animals would perhaps be coercion. This type of forceful control has, as far as I know, very little to do with how canids or equids exert dominance amongst themselves in established groups.

Rather, if we want to invoke the dominance concept as relevant with regards to how we care for animals, we might consider how to best distribute resources to a group of animals and ensure that we do that in a way so that those natural dominance interactions don’t escalate to outright aggression.

If space is limited, for instance, subordinate individuals can’t get displaced. And if they won’t move away from a resource, the dominant individual might start showing low-level or even escalated aggressive behaviour. Since dominance amongst animals is about resource distribution, it’s up to us as animal caregivers to distribute resources in a way that reduces the risk of escalated aggression.

Within applied ethology (using ethological principles to improve the quality of life of captive animals) we might do this along the following three lines of reasoning:

• Ensuring that there’s enough resources to go around for everyone and then some – and in different locations. Animals typically won’t share – we need to ensure that even the lowest ranking individuals can access those resources easily and out-of-sight of the dominant individuals, so feeding multiple animals simultaneously from the same bowl is often not a good idea. Dominance interactions tend to become exacerbated in captivity if space is limited (which it often is compared to how animals would group and distribute in the wild), so this is of tremendous importance.
• Ensuring that available space, including three-dimensional space where relevant, is big enough and accessible to the animals so that subordinate individuals can safely show displacement behaviour to avoid escalation of conflict.
• Another approach is allowing the dominant individual/s in a captive group to get exclusive access to a coveted toy or prized food, so that they can strut around and feel important, showing off. That will leave the less-coveted toys or food to the rest of the group. This can be a particularly effective way to reduce aggression around resources (because although they often don’t control the behaviour of other individuals, dominant animals often tend to control others’ access to coveted resources). And by offering them an irresistible resource, they’ll be so busy guarding that that all the other resources become available to the lower-ranking individuals of the group.

**saving some place here in case some benevolent reader has an image or video illustrating the strutting phenomenon that they’re willing to share…**

And now, I’ll get off that soapbox – let’s return to those labels.

Problem two: the Fundamental Attribution Error.

The second potential pitfall when using labels is that they might become detached from behaviour.

We might say ”he’s lazy” as if that explains every single behaviour the animal shows.

But behaviour is contextual, and using labels tends to sometimes remove us from the context. The label may also morph from being a shortcut when describing a specific behaviour in a specific context, to being used as a general personality trait.

And certainly, if we’re dealing with unwanted behaviour, we need to understand it – and to understand such behaviours, we must never lose sight of the context in which it occurs.

But what happens instead is that the label is seen as a trait, part of the animal’s personality. And so many people who work as behavioural consultants often see pet owners using that label almost as an excuse.

As if it absolves them of the responsibility of addressing the problem behaviour.

 ”Nothing I can do about that, he’s so dominant!” they may shrug and say, rather than realizing how a specific context triggers the unwanted behaviour in a particular situation – or how they might be contributing to the issue.

And here’s the thing: how we talk about the problem will influence how we solve the problem. And by labelling, we risk stopping to look for the actual cause.

In other words, those labels risk getting in the way of finding solutions.

We call this particular problem the fundamental attribution error. The tendency to under-estimate the importance of context in explaining an individual’s behavior, while over-emphasizing personality-based explanations.

So by labelling, one of the problems is that we risk the fundamental attribution error.

And of course, we do this with children too, all the time. Children are labelled bad, mad or inad, as in inadequate. Rather than understanding the importance of context, many frustrated parents tend to see their unruly behaviour as an unchangeable personality trait.

This leads us to the third problem.

Problem three: the Pygmalion / Rosenthal effect.

The self-fulfilling prophesy.

If we immediately label what we think we’re observing, that label will bias the observation; a self-fulfilling prophecy.

We will see the things that confirm the label, and miss or ignore the things that don’t confirm it.

This is a huge one, people.

Incidentally, it’s also one reason why in certain areas of science, scientific studies need to be blinded, and also ideally include a placebo treatment.

Why? Because the moment that the researcher or the participant of the study know that they’re getting a specific treatment, they subconsciously expect a specific outcome. And that expectation will not only impact the outcome through the intriguing powers of the placebo effect, but the interpretation of the results will also be skewed.

But if the study is blinded (as in, the researchers or participants don’t know who is getting which treatment), their observations will be more astute. And we include control groups also, to be able to tease out the effect of placebo alone as compared to the effect of placebo in combination with the actual treatment under study.

So, if we need to resolve some type of unwanted behaviour in the animals in our care, it’s useful to ditch the labels while exploring what’s going on – in order to remain unbiased during observations and reducing the risk of proverbially barking up the wrong tree and implementing the wrong solution.

For instance, some people may label their dog as dominant because he walks through the door first, and when he starts growling when they try to sit down next to him on the sofa, that confirms how dominant he is in their minds. When in fact, he has arthritis and the movement of the cushions as they plop down next to him is painful – hence the growling.

Oh, and incidentally, I have an extensive online course that goes into great detail on the topic of Preventing and Resolving Unwanted Behaviour in Animals, if you’re interested in learning more.

Problem four: Mis-Labelling

The fourth potential pitfall of labelling is inadvertently ascribing the wrong label to a behaviour that in fact belongs to different context, and should wear a different label.

Perhaps we see a snarling dog, for instance.

It might be a dominance-related behaviour; snarling could be part of an aggressive display.

Then again, the animal might be engaged in rough-and-tumble play; yes, snarling can occur during play in dogs, too.

So, we might see snarling, and label the dog as showing “aggression”, when in fact he’s playing.

In fact, I think it’s fair to say that many people mistake play for aggression. The play repertoire contains elements from the aggressive repertoire, but those responses are shown in a different order, they typically don’t escalate, and competent players often show self-handicap when faced with a smaller opponent, as well as role reversal – the players typically take turns being “victim” and “aggressor” – unlike real aggression. There’s also often meta-signals communicating the intent to play.

And we might handle a playful animal, an animal that is showing affiliative (friendly) behaviour, quite differently from how we’d handle an animal showing agonistic (aggressive or dominance-related) behaviour.

In other words, mis-labelling is a risk, and we might sometimes be better off simply describing the behaviours and the context in which they occur, rather than risk attaching the wrong label to the behaviour.

The Behaviour Analytic Approach

For the above reasons, a behaviour analytic approach (looking closely at the immediate and preceding context surrounding the unwanted behaviour; describing the responses that the animal is showing without labelling them; and identifying the reinforcers that maintain the unwanted behaviour – or the punishers preventing the desired behaviour) has great value when it comes to understanding how problem behaviour is maintained.

This approach reduces the risk of all these four pitfalls.

And indeed, every single behaviour analyst I’ve met has had a stark aversion to labels.

Since they typically work as behavioural consultants and help people resolve unwanted behaviours, they’ve seen these how their clients fall prey to these pitfalls, and how these four labelling problems get in the way of finding the solution – or might have even contributed to, or exacerbated, the problem.

Another argument against labelling is often brought forth by behaviour analysts.

The “Circular reasoning” argument

Here, we’re entering a territory where I believe that labels have been taking undeserved criticism.

Some people add another problem category to the four listed above, saying that “using labels is circular reasoning that is not scientifically verifiable.”

I disagree.

Here’s how the argument typically goes, through the example of an imaginary conversation:

”My dog is dominant.”

“How do you know?”

“Because he bites.”

“Why does he bite?”

“Because he’s dominant”.

And in one fell swoop, some people dismiss all labels – as if they were all circular. And while I don’t doubt that they have encountered this type of circular reasoning, they’re committing something akin to the fundamental attribution error by assuming that all labels are circular – rather than just some.

So, that particular argument is a gross misrepresentation of the ethological use of labels.

To reiterate, within ethology we use labels as a shorthand for a group of observable behaviours in a set of contexts with a certain category of outcomes. So, the same conversation between two ethologists might be:

”My dog is dominant over your dog.”

”How do you know?”

“Because he has priority of access to resource a, b, and c when they interact”.

What’s not ”scientifically verifiable” about that?

And how is that even circular?

If a client says “my dog is dominant”, we might choose to avoid labels altogether, or we might discuss what dominance really means and whether the observed behaviours would categorize under that label or not – and mention the potential pitfalls of using labels. Or – we might find a more useful label (keep reading for suggestions).

In my mind, labels have an undeservedly bad reputation amongst behaviour consultants, and I’m thinking that it’s most likely because their most vocal detractors don’t know how useful they might be in certain situations – or they’ve only seen them used badly (“my dog is so stubborn”).

And I don’t think we should judge the usefulness of a term or approach on the bad examples, neither should we unquestioningly accept the criticism of the label concept from people who don’t know how to use it.

That would be like taking someone who says ”I’ve tried that positive reinforcement thing, it doesn’t work” at his word..!

Another concern is that the potential benefits of labels get completely lost; I can’t count the number of times I’ve heard people completely dismiss them, saying things akin to “labels are for cupboards” and bending over backwards to avoid inadvertently using any type of label.

I’d much rather we had a discussion about useful versus harmful labels, instead of the “all labels must be avoided at all costs” stance that I feel permeates much of the behaviour analytical and animal training community.

Some behaviour consultants may use the labels “distance-increasing” or “distance-decreasing” to describe behaviours, in order to help their client better understand the nature of the problem.

And while I can see how that’s a useful way to reframe reinforcers and punishers, as well as using labels that won’t offend most hard-core behaviour analysts, I still find it lacking in information though. Take “distance-increasing” for instance, that label might typically involve either fear-related or aggression-related emotional states. And I would expect the physiological state, perception, decision-making and behaviours to be different for those two emotional / mood states – and perhaps the interventions would be different too, at least if the aggression is based in frustration rather than fear, perhaps. So I would explore using emotion labels instead.

Mind shift: Intentionally using positive labels and emotion labels

To wrap up, let’s dive into other arguments to explore the intentional use of labels – besides the ethological perspective.

For instance, labels may be negative (“my dog is stupid”), positive (“my dog is clever”), or toxic, based on unfulfilled expectations (“my dog really should be better behaved by now”).

Most of the arguments that I’ve seen against using labels involve examples where the label has a negative connotation (lazy, stupid, mean, stubborn). These negative labels carry emotional load; we think about the animal with negative connotations, falling prey to the Pygmalion effect and the Fundamental Attribution Error.

Removing the negative label and just describing the behaviour removes some of those negative connotations – and makes us more neutral observers. This is what many behaviour consultants suggest doing when dealing with unwanted behaviour in the animals in our care.

But what about adding a positive label – wouldn’t it do the opposite: add positive connotations, making us actively feel good about, and more invested in, the animal?

While we might possibly expect too much of an individual that we label as “smart”, these positive labels might contribute to our attachment to the individual and carry us through some difficult training. If we believe that our animal is smart, we will persist even if those first attempts fail – whereas if we thought the animal were stupid, we might just give up.

In other words, we could intentionally use positive labels to trigger the Pygmalion effect: if we expect the animal to change his behaviour for the better, we will tend to actually notice those instances when the animal is making good choices – and be less likely to notice the poor choices.

We might start by helping to reframe the animal’s personality in the clients’ minds by asking “what do you most love about the animal?” – this would be a way to find those positive labels and trigger the Pygmalion effect in more beneficial ways.

Another type of label that I think is useful are emotion labels. Labelling the animal as “fearful” or “anxious” may help us identify which emotional needs are not being met, and will give us ideas of how to help bring about a change in the underlying mood state – very often helping the pessimistic animal become more optimistic (more labels!) is all that is needed in order for the animal to stop showing unwanted behaviour.

To summarize, rather than the sweeping blanket statement often used by behaviour analysts along the lines of “labels won’t help us resolve the unwanted behaviour”, I would suggest to amend that to “negative labels won’t help us resolve the unwanted behaviour – but ethological labels, emotion labels or positive labels might, in some cases” – while keeping in mind the four risks involved when using said labels.

So, these are my current thoughts on this topic. And now, over to you: do you agree? Have you seen the four problems – or some other problem not mentioned here? Have you used ethological, emotional or positive labels, and if so, when and how – and to what effect?

***

I write the occasional blog post and give online masterclasses, courses and webinars, all on the topic of animal behaviour, learning and wellbeing. If you’re interested in hearing about what I’m up to, sign up below and I’ll keep you in the loop!

Selected references:

Banerjee, Arunita, and Anindita Bhadra. “Time-activity budget of urban-adapted free-ranging dogs.” Acta Ethologica 25, no. 1 (2022): 33-42.

Benedetti, Fabrizio. “Placebo effects: understanding the mechanisms in health and disease.” Oxford University Press, USA, 2021.

Jensen, Per. “The ethology of domestic animals 2nd edition: an introductory text.” Oxfordshire, CAB International (2009).

Odendaal, J. S. J. (1997). ”An ethological approach to the problem of dogs digging holes.” Applied Animal Behaviour Science, 52(3-4), 299-305.

Paukner, Annika, and Stephen J. Suomi. “Sex differences in play behavior in juvenile tufted capuchin monkeys (Cebus apella).” Primates 49 (2008): 288-291.

Raudenbush, Stephen W. (1984). “Magnitude of teacher expectancy effects on pupil IQ as a function of the credibility of expectancy induction: A synthesis of findings from 18 experiments”. Journal of Educational Psychology. 76: 85–97.

Stafford, Richard, Anne E. Goodenough, Kathy Slater, William S. Carpenter, Laura Collins, Heather Cruickshank, Sarah Downing et al. “Inferential and visual analysis of ethogram data using multivariate techniques.” Animal Behaviour 83, no. 2 (2011): 563-569.

Young, Robert J. “The importance of food presentation for animal welfare and conservation.” Proceedings of the Nutrition Society 56, no. 3 (1997): 1095-1104. D